chrishkchris commented on a change in pull request #694:
URL: https://github.com/apache/singa/pull/694#discussion_r427863586
##########
File path: include/singa/core/device.h
##########
@@ -51,6 +51,17 @@ using std::vector;
namespace singa {
+/// Method used by time profiling
+/// ProfilingMode == 0 (default) -> Chrono
+/// ProfilingMode == 1 -> Event (e.g. cudaEvent)
+enum ProfilingMode { useChrono, useEvent };
Review comment:
yes, cpu can use chrono only, gpu can use either 1. chrono or 2.
cudaevent
Time Profiling methods in NVIDIA documentation:
1. Chrono:
https://docs.nvidia.com/deeplearning/sdk/tensorrt-best-practices/index.html#cpu-timing
2. CudaEvent:
https://docs.nvidia.com/deeplearning/sdk/tensorrt-best-practices/index.html#cuda-events
I have implemented both for GPU
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
[email protected]
